Collaboration
Excited to collaborate!
I have a variety of topics and ideas that I am interested in and would love to collaborate on! As mentioned earlier, I am broadly interested in search algorithms, combining them with machine learning/reinforcement learning, and motion planning.
Some example medium-levels ideas I would be happy to discuss:
- MAPF or multi-agent motion planning for non-gridworld systems including:
- Continuous space planning
- Incorporating kinematics and dynamics constraints
- Heterogenuous multi-agent team compositions
- Real-Time or windowed planning algorithms for multi-agent systems with gaurantees
- Decentralized MAPF and multi-agent motion planning algorithms
- Meta-reasoning in heuristic search algorithms
- Search algorithms for LLMs
- Single-agent heuristic search algorithms with/for RL
- Foundation models for MAPF
- Cooperative MAPF
Some specific project ideas where I have concrete algorithms which someone could work on (e.g. graduate student or motivated undegraduate):
- Extensions to PIBT
- Extensions to Windowed MAPF with Completeness Gaurantees:
- Continuous space or non-gridworld systems with local collision avoidance (e.g. swarms with local MPC or RL controllers)
- Lifelong MAPF (as opposed to single-shot MAPF)
- Extensions to Learning Local Heuristics (e.g., for more complex domains like manipulation)
- MAPF with limited cooperative actions
- Single-agent heuristic search algorithms with/for RL